HAL Id: hal-03008915
https://hal.archives-ouvertes.fr/hal-03008915
Submitted on 6 Dec 2020
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
synthesis at arrested forks by promoting SUMO removal
Karol Kramarz, Kamila Schirmeisen, Virginie Boucherit, Anissia Ait Saada,
Claire Lovo, Benoît Palancade, Catherine Freudenreich, Sarah Lambert
To cite this version:
Karol Kramarz, Kamila Schirmeisen, Virginie Boucherit, Anissia Ait Saada, Claire Lovo, et al.. The
nuclear pore primes recombination-dependent DNA synthesis at arrested forks by promoting SUMO
removal. Nature Communications, Nature Publishing Group, 2020, 11 (1),
�10.1038/s41467-020-19516-z�. �hal-03008915�
The nuclear pore primes recombination-dependent
DNA synthesis at arrested forks by promoting
SUMO removal
Karol Kramarz
1,2,3
, Kamila Schirmeisen
1,2,3
, Virginie Boucherit
1,2,3
, Anissia Ait Saada
1,2,3
,
Claire Lovo
1,2
, Benoit Palancade
4
, Catherine Freudenreich
5
& Sarah A. E. Lambert
1,2,3
✉
Nuclear Pore complexes (NPCs) act as docking sites to anchor particular DNA lesions
facilitating DNA repair by elusive mechanisms. Using replication fork barriers in
fission yeast,
we report that relocation of arrested forks to NPCs occurred after Rad51 loading and its
enzymatic activity. The E3 SUMO ligase Pli1 acts at arrested forks to safeguard integrity of
nascent strands and generates poly-SUMOylation which promote relocation to NPCs but
impede the resumption of DNA synthesis by homologous recombination (HR). Anchorage
to NPCs allows SUMO removal by the SENP SUMO protease Ulp1 and the proteasome,
promoting timely resumption of DNA synthesis. Preventing Pli1-mediated SUMO chains
was suf
ficient to bypass the need for anchorage to NPCs and the inhibitory effect of
poly-SUMOylation on HR-mediated DNA synthesis. Our work establishes a novel spatial control of
Recombination-Dependent Replication (RDR) at a unique sequence that is distinct from
mechanisms engaged at collapsed-forks and breaks within repeated sequences.
https://doi.org/10.1038/s41467-020-19516-z
OPEN
1Institut Curie, PSL Research University, UMR3348, F-91405 Orsay, France.2CNRS UMR3348“Genome integrity, RNA and Cancer”, “Equipe labellisée
LIGUE 2020”, F-91405 Orsay, France.3University Paris Sud, Paris-Saclay University, UMR3348, F-91405 Orsay, France.4Université de Paris, CNRS, Institut
Jacques Monod, F-75006 Paris, France.5Department of Biology, Tufts University, Medford, MA 02155, USA. ✉email:sarah.lambert@curie.fr
123456789
F
laws in the DNA replication process, known as replication
stress, lead to fragile replication fork structures prone to
chromosomal rearrangement and mutation, contributing to
human diseases including cancer
1,2. The resolution of replication
stress occurs within a compartmentalized nucleus. How the
dis-tinct nuclear compartments operate to ensure faithful resolution
of replication stress is far from understood.
The completion of DNA replication is continuously threatened
by numerous obstacles. Replication obstacles hinder fork elongation
and occasionally cause dysfunctional forks, deprived of their
repli-cation competence
3. Replication-based pathways have evolved to
ensure DNA replication completion and avoid genome instability.
Dysfunctional forks are either rescued by opposite forks or, if a
converging fork is not available in a timely manner, restarted and
repaired. Homologous recombination (HR) is a ubiquitous DNA
repair pathway involved in the repair of double strand breaks
(DSBs), and in the protection and restart of dysfunctional forks
3.
This last pathway is referred to as recombination-dependent
replication (RDR), a DSB-free mechanism allowing efficient
fork-restart. The pivotal HR protein is the recombinase Rad51 that is
loaded onto single-stranded DNA (ssDNA) with the help of its
loader Rad52 in yeast. At compromised forks, the combined action
of nucleases promotes the resection of newly replicated strands to
generate ssDNA gaps and the subsequent loading of Rad51
4. Then,
the strand exchange activity of Rad51 builds a particular DNA
structure, called a D-loop, from which DNA synthesis is primed
allowing fork-restart
5,6. A feature of RDR is its mutagenic DNA
synthesis prone to chromosomal rearrangements
7–10. How the
subsequent steps of RDR are spatially segregated within the nuclear
architecture is unknown.
The nuclear periphery (NP) constitutes a boundary between
the nucleus and cytoplasm and is formed of a double membrane
nuclear envelope (NE) and multiple nuclear pore complexes
(NPCs)
11. NPCs are highly conserved macromolecular structures,
composed of multiple copies of 30 different nucleoporins, most of
which associate in stable sub-complexes
12–14. A central channel
(referred to as the core of NPCs) allows macromolecule exchange
between the cytoplasm and the nucleus. The largest NPC
sub-complex is the Y-shaped mammalian Nup107-Nup160 sub-complex
(called Nup84 complex in budding yeast), located both at the
cytoplasmic and nuclear side
15.
In budding yeast, DNA lesions (persistent DSBs, eroded
telo-meres, and collapsed forks) shift to the NP to associate with two
distinct perinuclear anchorage sites: either the inner nuclear
membrane SUN protein Mps3 or NPCs (extensively reviewed in
ref.
16). DSB-NPC association occurs in all cell cycle phases
whereas DSB-Mps3 association is restricted to S/G2 cells.
Relo-cation of DSBs to either Mps3 or the NPC requires distinct
sig-naling mechanisms to promote distinct DNA damage survival
pathways
17–24. The
fission yeast homologue of Mps3, Sad1, was
shown to co-localize with DSBs, indicating an evolutionarily
conserved role of the NE in DSB repair
25.
Anchoring of DNA lesions to NPCs requires SUMOylation
events, a type of post-translational modification
17,20,22,23,26.
The SUMO (Small Ubiquitin-like Modifier) particle is
cova-lently bound to lysines of target proteins by the joint action of
SUMO-activating (E1) and -conjugating (E2) enzymes, a
pro-cess enhanced by SUMO E3 ligases
27,28. Persistent DNA
damage and eroded telomeres are subject to SUMOylation
waves that target DNA repair factors
29,30. SUMOylated
pro-teins are key substrates for the SUMO Targeted Ubiquitin
Ligase (STUbL) family of E3 ubiquitin ligases such as the yeast
Slx8-Slx5 and human RNF4, that target DNA lesions to
NPCs
17,20,22,23,26,31–33. SUMOylated proteins can undergo
degradation or direct SUMO removal by SENP proteases, which
are spatially segregated within the nucleus
34. In yeasts, the
SENP protease Ulp1 is constitutively attached to NPCs, whereas
Ulp2 is found in the nucleoplasm
35,36.
The NPC has emerged as a central player in the maintenance of
genome integrity
37,38. Mutations in the budding yeast Nup84
complex lead to a defective DNA repair and replication stress
response
11,17,36,39–41. The outcome of relocation of damage is
often deduced from the phenotypes arising from the ablation of
anchorage sites at NPCs. Budding yeast NPCs favor the repair of
DSBs by Break Induced Replication (BIR)
20,42. Eroded telomeres
relocate to NPCs in a SUMO-dependent manner to allow
recombination-mediated elongation of telomeres, generating type
II survivors
23. A failure in anchoring forks stalled at expanded
CAG repeats leads to chromosomal fragility of CAG tracts
22.
Also, delocalization of Ulp1 caused by mutations in the Nup84
complex results in DNA damage sensitivity
36but how
Ulp1-associated NPCs safeguard genome integrity is poorly
under-stood. In eukaryotes, breaks within repeated sequences
(Hetero-chromatin, rDNA) shift away from their chromatin environment,
in a SUMO-dependent manner, to allow Rad51 loading and the
completion of HR repair
26,43–46. Thus, an emerging scenario
suggests that NPCs are involved in both SUMO homeostasis and
anchoring of DNA lesions to spatially segregate DNA repair
events and avoid inappropriate HR repair. However, failures in
uncoupling SUMO homeostasis from anchorage did not allow
interrogating the relative contributions of these two NPC
func-tions in maintaining genome integrity.
Using a site-specific replication fork barrier (RFB), we report
that DSB-free and dysfunctional forks relocate and anchor to
NPCs, in a poly-SUMO and STUbL-dependent manner, for the
time necessary to complete RDR. Relocation occurs after Rad51
binding and enzymatic activity, suggesting that D-loop
inter-mediates anchor to NPCs. We reveal a novel post-anchoring
function of NPCs in promoting the removal of SUMO chains by
Ulp1 and the proteasome. Indeed, the E3 SUMO ligase
Pli1 safeguards fork-integrity and generates SUMO chains that
trigger NPC anchorage but further limit the efficiency of
HR-mediated DNA synthesis. Selectively preventing Pli1-dependent
SUMO chains is sufficient to bypass the need for NPC anchorage
in promoting HR-mediated DNA synthesis. We uncovered a
novel SUMO-based regulation that spatially segregates the
sub-sequent steps of RDR and that is distinct from mechanisms
engaged at DSBs and collapsed forks within repeated sequences.
Results
To investigate the spatial regulation of RDR, we exploited the
RTS1-RFB that allows a single replisome to be blocked in a polar
manner at a defined locus on S. pombe chromosome III (Fig.
1
a).
The RFB activity is mediated by the RTS1-bound protein Rtf1
whose expression is repressed in the presence of thiamine
47.
Forks arrested at the RFB become dysfunctional and are rescued
by opposite forks or, if not available in a timely manner, restarted;
both pathways require the binding of Rad51 to the active RFB
6.
Replication fork restart occurs by RDR within
∼20 min and is
initiated by an end-resection machinery to generate ssDNA gaps
onto which RPA, Rad52, and Rad51 are loaded
4,5,48,49. RDR is
associated with a non-processive DNA synthesis liable to
repli-cation slippage and GCRs, during which both strands are
syn-thetized by Polymerase delta, making the progression of restarted
forks likely insensitive to the RFB
7,49.
Dysfunctional forks associate with NPCs for
∼20 min during
S-phase. To follow the sub-nuclear location of the active RFB in
living cells, we employed a marked RFB visualized by
LacO-bound mCherry-LacI foci in yeast expressing the endogenous
tagged Npp106-GFP, a component of the inner ring complex of
NPCs (Fig.
1
a, b)
6. The shape of the nucleus in S and G2-phase
cells was often irregular, preventing us to apply a classical zoning
approach
17to assign the nuclear positioning of the LacO-marked
RFB. Instead, we monitored co-localization between the NP and
the LacO-marked RFB (Fig.
1
b, c). When the RFB was inactive
(RFB OFF) or absent from the ura4
+locus (no RFB, Fig.
1
a),
LacI-foci co-localized with the NP in
∼45% of both S and
G2-phase cells (Fig.
1
c). Upon activation of the RFB (RFB ON), the
LacO-marked RFB was located more frequently at the NP in
S-phase cells,
∼70% of the time, but not in G2 cells. Thus, forks
0.6 80 70 60 50 40 30 20 Co-localization in S-phase % 10 0 ori ori oriLacl-bound LacO repeats ori RFB t-LacO-ura4<ori t-LacO-ura4-ori LacO-marked RFB ura4+ ura4+ cen3 Main replication direction
a
d
f
g
h
e
b
c
0.5 MSD ( μ m 2) MSD ( μ m 2) 0.4 0.3 0.2 0.1 0 0 50 100 S phase RFB OFF Rc = 0.74 μm n = 20 RFB ON Rc = 0.57 μm n = 20 RFB ON RFB ON RFB OFF RFB ON RFB OFF Npp106-GFP n = 4 Man1-GFP Sad1-GFP n = 3 n = 3 RFB ON RFB OFF RFB ON RFB OFF RFB OFF no RFB no RFB RFB ON RFB ON RFB OFF RFB OFF no RFB RFB ON no RFB ns RFB OFF Rc = 0.71 μm n = 13 RFB ON Rc = 0.69 μm n = 13 G2 phase G2 phase S phase p = 0.027 p = 0.0002 p = 0.0002 p < 0.0001 p = 0.0182 p = 0.0173 p = 0.0003 150 200 Npp106-GFP Δt (s) 0 0 0 5 10 15 20 25 30 min Time (min) 30 50 100 150 200 Cell # 7 Cells ( n = 10) Cells ( n = 10) Cells ( n = 10) Cell # 8 Cell # 1 –110 110 400 ade6 bp –110 110 400 cen ade6 bp –110 110 400 cen ade6 bp Δt (s) 0.6 2.5 2.0 Fold enrichment Fold enrichment Fold enrichmentAverage time of co-localization
in min in septated cells
1.5 1.0 0.5 0.0 2.5 2.0 1.5 1.0 0.5 0.0 50 30 25 20 15 10 5 0 30 10 2 1 0 0.5 0.4 0.3 0.2 0.1 5 μ m 5 μm 1 min 5 μ m 5 μ m 0
arrested by a DNA-bound protein complex transiently relocate to
the NP in S-phase cells.
To examine if the dynamics of the active RFB changes with NP
enrichment, we monitored the mobility of the GFP-LacI focus by
single-particle tracking (SPT) in living cells (Supplementary Fig. 1a)
and calculated the range of nuclear volume explored by the
LacO-marked RFB by mean square displacement (MSD) analysis (Fig.
1
d)
as reported for other types of damage
50. Upon RFB activation, the
overall mobility of the LacO-marked RFB decreased, exclusively in
S-phase cells, compared to the RFB OFF control. The radius of
constraint (Rc, radius of maximum volume of particle movement)
in the OFF condition was significantly higher than the one obtained
in the ON condition in S phase cells (p < 0.05) while no significant
difference was detected in G2 cells, indicating that dysfunctional
forks exhibit a reduced mobility in S-phase, consistent with an
anchorage to a perinuclear structure. To identify the anchorage site,
we performed Chromatin Immunoprecipitation (ChIP)
experi-ments against Npp106-GFP, Sad1-GFP (the Mps3 orthologue) and
Man1-GFP (a Lap-Emerin-Man domain protein of the inner
nuclear envelope) to test their binding to the RFB. Man1 and Sad1
were found enriched at centromeres, as reported
51,52, but not at the
active RFB (Fig.
1
e). Npp106-GFP was significantly enriched at the
active RFB, indicating that NPCs are acting as anchorage sites as
reported for extended CAG repeats
22. In these experiments, we
used strains devoid of the nearby LacO array to ensure the binding
of NP components to the active RFB is not a consequence of
proximal LacO arrays that may influence sub-nuclear positioning.
To investigate the dynamics of the association of the RFB with
the NP in single cell, we performed time-lapse microscopy for 30
min to build up kymographs over time (See
“Methods” and
Supplementary Fig. 1a). The analysis of 10 individual S-phase nuclei
showed short and intermittent co-localizations between the NP and
the unstressed locus (RFB OFF and no RFB controls), indicating
transient and dynamic interactions (Fig.
1
f, g and Supplementary
Fig. 1b–d). The average time of co-localization was ∼10 min
(Fig.
1
h). Consistent with an anchorage to NPCs, the active RFB
co-localized with the NP in a less sporadic manner, with interactions
lasting for most of the acquisition time in the majority of S-phase
cells analyzed. The average time of co-localization was
∼20 min
(Fig.
1
h), and correlated with the time needed to restart replication
forks
48,49. We conclude that dysfunctional forks transiently anchor
to NPCs in S-phase, for a time that coincides with the time needed
to complete RDR.
Relocation to NPCs requires Rad51 loading and enzymatic
activity. Collapsed forks but not stalled forks associate to
NPCs
17,22. Because the exact nature of DNA structures
under-lying collapsed versus stalled forks remains debated, we addressed
the role of fork processing in anchoring the RFB to NPCs. The
resection of nascent strands at arrested forks primes RDR. It
occurs as a two-step process: a short-range resection by
MRN-Ctp1 that generates
∼110 bp sized gaps obligatory for replication
restart followed by an Exo1-mediated long-range resection
5. One
role of MRN-Ctp1 is to remove the heterodimer KU from
dys-functional forks to overcome its anti-resection activity.
Conse-quently, the lack of KU results in extensive fork-resection. We
observed a lack of correlation between the extent of fork-resection
and the capacity of the active RFB to shift to the NP and bind to
NPCs (Fig.
2
a, b, see Supplementary Fig. 2 for location in
G2-phase). Instead, we noticed that RFB relocation was abrogated in
mutants exhibiting a delay in replication restart (i.e. rad50Δ,
ctp1Δ and pku70
5) raising the possibility that
replication/recom-bination intermediates formed during RDR trigger relocation to
NPCs. Consistent with this, Rad51 and Rad52 were necessary to
shift the active RFB to the NP (Fig.
2
c and Supplementary Fig. 2).
Rad51 promotes replication restart at arrested forks and protects
them from uncontrolled end-resection to facilitate merging with
opposite forks. To distinguish between these two Rad51
func-tions, we analyzed the rad51-II3A mutant that binds DNA to
protect forks but is unable to facilitate restart because of its
defective strand exchange activity
6. The active RFB did not shift
to the NP nor bind to NPCs in rad51-II3A cells (Fig.
2
b, c and
Supplementary Fig. 2), reinforcing the notion that relocation
occurs after fork remodeling by Rad51 enzymatic activity. Since
MRN-Ctp1 is active in rad51-II3A cells, we propose that
short-range resection mediated by MRN-Ctp1 is necessary but not
sufficient to shift arrested forks to NPCs and that building
Rad51-mediated joint-molecules at arrested forks is necessary for stable
association with NPCs.
RDR and anchorage, but not fork-integrity, are impaired by
the loss of the Slx5-Slx8 STUbL pathway. Depending on the
nature of DNA lesions, the S. pombe Slx8 STUbL either
sup-presses or promotes genome instability
53. Also, Slx8 prevents
uncontrolled HR at the constitutive RTS1-RFB
54. Thus, it was
worthwhile to address the role of SUMO and Slx8 activity in the
spatial regulation of RDR. SUMO (encoded by the
non-essential S. pombe gene pmt3
+) was necessary to shift the
active RFB to the NP in S-phase (Fig.
3
a and Supplementary
Fig. 3a). In the temperature-sensitive slx8-29 mutated strain
55,
the active RFB did not shift to the NP at 32 °C (Fig.
3
a
and Supplementary Fig. 3a) and MSD analysis showed an
Fig. 1 The activeRTS1-RFB transiently relocates to NPCs in S-phase. a Scheme of the LacO-marked RTS1-RFB (purple) integrated at the ura4+locus
(green, t-LacO-ura4 < ori) or not (t-LacO-ura4-ori). Cen3: centromere position. LacO arrays (red) bound by mCherry-LacI (ellipses) are integrated ∼7 kb
away from ura4+. When Rtf1 is expressed (RFB ON, 24 h induction for cell imaging experiments) and binds to RTS1, 90% of forks moving from cen3 to t are
blocked.b Example of co-localization between Npp106-GFP and the LacO-marked RFB. Mono-nucleated cells and septated bi-nucleated cells correspond to
G2 and S-phase cells, respectively. Arrows indicate co-localization events.c Quantification of co-localization events in indicated conditions: t-LacO-ura4-ori,
Rtf1 expressed (no RFB), t-LacO-ura4 < ori, Rtf1 repressed (RFB OFF) and t-LacO-ura4 < ori, Rtf1 expressed (RFB ON). n = 250 cells in both S and G2 phase.
Two-sided Fisher’s exact test was used for group comparison to determine the p value (ns non-significant). Dots represent values from two independent
biological experiments.d The mobility of the RFB in OFF and ON conditions is presented as a mean square displacement (MSD) over the indicated time
interval (Δt) for n independent cells. Rcradius of constraint. p value was calculated as a one sided t-test based on MSD curves. Black bars correspond to
standard error of the mean (SEM).e Binding of the RFB to Npp106-GFP (top), Man1-GFP (middle) and Sad1-GFP (bottom) analyzed by ChIP-qPCR.
Distances from the RFB are presented in bp. A centromere locus, known to interact with Man1 and Sad1 was used as a positive control. Primers targeting ade6 gene were used as unrelated control locus. Values are mean of n independent biological repeats, with standard deviation (SD) as error bars. p value was calculated using two-sided t-test. f Representative kymographs over 30 min of single S phase nucleus in indicated conditions. Green and red signals correspond to the Npp106-GFP marked nuclear periphery and the LacO-marked RFB, respectively. g Co-localization time from the analysis of kymographs
in indicated conditions. Each line corresponds to an individual S-phase nucleus. Ten cells per conditions were analyzed.h Average co-localization
time obtained fromf. Each dot represents one sample, red bar indicate the mean from 10 independent S-phase cells ± SD. p value was calculated using
increased mobility of the active RFB (Fig.
3
b), indicating a lack
of anchorage to NPCs when Slx8 is not functional. At
per-missive temperature (25 °C), the slx8-29 mutated strain behaved
as WT control (Figs.
3
b and
1
d). Rfp1 and Rfp2 are two
orthologues of Slx5 and they form two independent
hetero-dimers with Slx8
31. The active RFB did not shift to the NP in
the absence of either Rfp1 or Rfp2 (Fig.
3
a and Supplementary
Fig. 3a), reinforcing the notion that the Slx8 STUbL anchors
arrested forks to NPCs.
To address the consequences of this lack of relocation, we
investigated the efficiency of RDR. HR-mediated fork restart is
associated with a non-processive DNA synthesis liable to
replication slippage (RS). We developed genetic assays to monitor
RFB-induced RS, based on the restoration of a functional ura4
+gene to select for Ura
+cells (Fig.
3
c and details in the legend)
7.
The frequency of Ura
+reversion is used as readout of the
frequency at which the ura4-sd20 allele is replicated by a restarted
fork in the cell population. At 32 °C, the frequency of
80
a
b
c
p = 0.0005 p = 0.0036 p = 0.0196 p = 0.0021 p = 0.0017 n = 4 n = 4 n = 3 n = 3 n = 3 p = 0.0162 p = 0.0002 p = 0.0008 p = 0.0064 ns ns ns ns ns ns ns ns 70 60 50 40 Co-localization in S-phase % 30 20 10 WT WT WT pku70Δ pku70 exo1Δ ctp1Δ rad50Δ rad51Δ rad52Δ rad50 rad51-II3A rad51-II3A exo1 rad50Δ exo1Δ 0 RFB ON RFB ON RFB OFF RFB OFF no RFB RFB ON RFB OFF no RFB 2.5 2.0 Fold enrichment 1.5 1.0 0.5 0.0 2.5 2.0 Fold enrichment 1.5 1.0 0.5 0.0 2.5 80 70 60 50 40 30 20 10 0 2.0 Fold enrichment 1.5 1.0 0.5 0.0 2.5 2.0 Fold enrichment 1.5 1.0 0.5 0.0 2.5 2.0 Fold enrichment Co-localization in S-phase % 1.5 1.0 0.5 0.0 –110 110 400 –110 110 RFB 400 bp ade6 –110 110 400 ade6 –110 110 400 ade6 –110 110 400 ade6 –110 110 400 ade6Fig. 2 Relocation to NPCs requires Rad51 enzymatic activity. a Co-localization events in S-phase cells in indicated conditions and strains, as described on
Fig.1b, c. p value was calculated by Fisher’s exact test for OFF and ON groups for each mutant and condition. In all, 200 cells were analyzed for each strain
and condition. Dots represent values obtained from two independent biological experiments. For each set of data, WT strain was analyzed alongside
mutants.b Binding of Npp106-GFP to the RFB in indicated strains. Upstream and downstream distances from the RFB are presented in bp (top). Primers
targeting ade6 gene were used as unrelated control locus. Values are mean of n independent biological repeats, with SD as error bars. p value was calculated using two-sided t-test. c Co-localization events in S-phase cells in indicated conditions and strains, as in a.
RFB-induced RS in slx8-29 cells was decreased by nearly 50%,
compared to WT (Fig.
3
d) indicating that Slx8 promotes RDR.
This defect was not caused by a less efficient Rad51 binding to the
active RFB (Fig.
3
e). Finally, we investigated the integrity of fork
arrested by the RFB. We analyzed replication intermediates by
bi-dimensional gel electrophoresis (2DGE) to examine the resection
of nascent strands at arrested forks (referred to as resected forks,
Fig.
3
f)
5. The lack of a functional Slx8 pathway (in slx8-29, rfp1Δ,
rfp2Δ or double mutants) did not impede or enhance the level of
resected forks (Fig.
3
f, g and Supplementary Fig. 3b, c). Hence,
the lack of Slx8-mediated anchorage to NPCs impedes
HR-mediated DNA synthesis downstream of fork-resection and
Rad51 loading, suggesting that the processing of SUMO
conjugates is necessary to complete RDR.
Nup132 promotes HR-dependent DNA synthesis in a
post-anchoring manner. To elucidate the mechanisms engaged at
NPCs, we focused on the two
fission yeast orthologues of
Nup133, a component of the Y-shaped Nup107-Nup160
com-plex: Nup132 that is the most abundant (∼3000 molecules/cell),
and localized at the nuclear side of NPCs, whereas Nup131 is less
expressed (∼200 molecules/cell) and is localized at the
cyto-plasmic side
56. Interestingly, nup132Δ cells, but not nup131Δ
cells, were sensitive to a broad range of replication-blocking
agents, including hydroxyurea (HU), but not to DSBs induced by
bleomycin or to UV-induced DNA damage (Fig.
4
a). A major
function of NPCs being the transport of macromolecules, we
further analyzed protein import and mRNA export in these
mutants. Neither the absence of Nup131 nor Nup132 affected
80a
b
c
f
g
d
e
70 60 50 40 Co-localization in S-phase % 30 20 10 0.8 80 60 40 20 0 3 2 1 0 3 0 5 10 15 20 25 2 1 0 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 80 70 60 50 40 30 20 10 0 WT WT WT WT WT 32 °C WT, 25 °C slx8-29, 32 °C slx8-29, 25 °C ade6 control locus 25°32°25°32 °C slx8-29 WT WT, 32 °C 25 °C 32 °C Resected forks % (relative to arrested forks)slx8-29 25 °C slx8-29 WT slx8-29 slx8-29, S phase, 25 °C slx8-29, S phase, 32°C slx8-29 32 °C pmt3 rfp1 rfp2 rfp1 rfp2 rfp1Δ rfp2Δ rfp1Δ rfp2Δ rfp1Δ rfp2Δ MSD ( μ m 2)
Rad51 enrichment (ratio ON/OFF)
Frequency of RFB-Induced Ura + reversion X 10 -5 MSD ( μ m 2) RFB OFF Rc = 0.72 μm n = 14 RFB ON Rc = 0.54 μm n = 11 RFB ON Rc = 0.9 μm n = 27 RFB OFF Rc = 0.69 μm n = 27 p = 0.045 p = 0.0082 0 50 100
spontaneous replication slippage
replication slippage occuring during restart
150 200 Δt (s) 0 50 100 Large Y Arrested fork Resected fork 93.6 ± 0.3 92 ± 0.2 Secondary signal Secondary signal 150 200Δt (s) RFB ON RFB ON RFB OFF RFB OFF no RFB t-ura4sd20-ori t-ura4sd20<ori p = 0.0005 p = 0.0003 p < 0.0001 p = 0.0021 ns ns ns ns ns ns ura4-sd20 ura4-sd20 RFB ori
ori ori ori
ori ori ori
ori cen3 cen3 n = 12 n = 3 n = 3 n = 3 n = 3 n = 12 n = 11 n = 8 -110 110 400 bp RFB
nuclear shape and protein import, but nup132Δ cells exhibited a
very mild defect in mRNA export (Supplementary Fig. 4) albeit
moderate when compared to the strong defect reported upon heat
shock
57.
We tested the role of Nup132 in the recovery from HU-stalled
forks. Strains were blocked in early S-phase by exposing
exponentially growing cells to HU for 4 hours and then released
into HU-free media. Flow cytometry analysis indicated that the WT
and nup131Δ strains reached a G2 DNA content within 45 min
after release whereas nup132Δ and nup131Δ nup132Δ cells
exhibited an additional 15 min delay (Supplementary Fig. 5a, left
panel). Chromosome analysis by Pulse Field Gel Electrophoresis
(PFGE) showed that HU treatment prevented chromosomes from
migrating into the gel because of the accumulation of replication
intermediates (Supplementary Fig. 5b). Sixty minutes after release,
WT chromosomes were able to migrate into the gel and their
intensity doubled 90 minutes after release, indicating that the WT
genome was fully duplicated and replication intermediates were
resolved (Supplementary Fig. 5b, c). In contrast, chromosomes from
nup132Δ cells showed a clear delay in their ability to migrate fully
into the gel. Even 90 minutes after release, chromosomes intensity
did not double, indicating that nup132Δ genome failed to be fully
duplicated because of an accumulation of unresolved replication
intermediates. Our data reveal a critical role for Nup132 in
promoting DNA replication upon transient fork stalling.
We asked if Nup132 and Nup131 are involved in RDR. We
detected a reduced frequency of RFB-induced RS only in the
absence of Nup132 and no further reduction was observed in the
double nup131Δ nup132Δ mutant (Fig.
4
b). This defect was not
correlated with a less efficient Rad51 binding to the active RFB
(Fig.
4
c), indicating that the early step of RDR, fork-resection and
Rad51 loading, are functional. The active RFB was enriched at the
NP in S-phase cells in the absence of either Nup131 or Nup132, but
not in the absence of both nucleoporins (Fig.
4
d and Supplementary
Fig. 2). Supporting this result, the active RFB bound properly to
NPCs in nup132Δ cells but not in the double nup131Δ nup132Δ
mutant by ChIP (Fig.
4
e). Thus, Nup132 is dispensable to anchor
remodeled forks to NPCs. However, the absence of both
nucleoporins may modify the NPC structure, making it inefficient
for anchoring. These data reveal a novel function for NPCs in
which Nup132 promotes HR-dependent DNA synthesis,
down-stream of Rad51 binding, in a post-anchoring manner.
HR-dependent DNA synthesis is non-processive, liable to
mutation, and GCR. We monitored the rate of RFB-induced
mutagenesis and GCR, including translocation and genome deletion
(Supplementary Fig. 6a, b for detailed explanations)
7. Briefly, we
selected ura4 loss events after RFB induction or not and analyzed
the events by PCR to discriminate between mutation, translocation,
and genomic deletion; all these events occur in an HR-dependent
manner. In WT cells, the induction of the RFB resulted in a 4.5, 10,
and 14-fold increase in the rate of mutagenesis, deletion, and
translocation, respectively (Supplementary Fig. 6c, d). The rate of
translocation and genomic deletion were unaffected in the absence
of Nup131 and Nup132, but RFB-induced mutagenesis was
abolished in nup131Δ and nup132Δ single mutants or in the
double mutant, indicating a role of both nucleoporins in promoting
mutagenic HR-mediated DNA synthesis. Altogether, our data
reveal a novel NPC function, via Nup132 and to a lesser extent
Nup131, in promoting HR-dependent DNA synthesis. The distinct
contribution of Nup131 and Nup132 to this pathway might reflect
their different localization within NPCs and/or their relative
abundance
56.
Pli1-dependent SUMO chains are toxic to HR-dependent DNA
synthesis. Our data indicate that anchoring to NPCs is not
suffi-cient to promote RDR, as exemplified in the nup132Δ mutant. In
the absence of Nup132, the SUMO deconjugating enzyme Ulp1 is
delocalized from NPCs and can no longer antagonize the PIAS
family E3 ligase Pli1 that promotes 90% of bulk SUMOylation and
SUMO chain formation. As a consequence, both Ulp1 and Pli1
expression are lowered, resulting in a low global SUMOylation
level
35. Surprisingly, the deletion of pli1 partly rescued the
sensi-tivity of nup132Δ cells to replication stress (Fig.
5
a), suggesting a
toxicity of Pli1 activity in the absence of Nup132. We asked if this
toxicity might also underlie the RDR defect. The active RFB did not
shift to the NP nor bound to NPCs in the absence of Pli1 (Fig.
5
b, c
and Supplementary Fig. 2). MSD analysis confirmed an absence of
reduced mobility of the active RFB and thus a lack of anchorage in
pli1Δ cells (Fig.
5
d). However, the lack of Pli1 did not affect
RFB-induced RS (Fig.
5
e), indicating that RDR is fully completed without
anchorage to NPCs when Pli1 is absent. Interestingly, the lack of
Pli1 partly rescued the defect in RFB-induced RS of nup132Δ
cells (Fig.
5
e), even though the active RFB was still unable to bind
NPCs (Fig.
5
b, c). A similar rescue was observed in slx8-29 pli1Δ
cells (Fig.
5
e), consistent with Pli1 causing genome instability in the
absence of STUbL activity
54,55. Of note, the deletion of pli1 did not
rescue the mRNA export defect of nup132Δ cells, showing that the
role of Nup132 in promoting RDR and mRNA export are
uncou-pled (Supplementary Fig. 4d, e). Thus, Pli1 activity is necessary to
anchor arrested forks to NPCs but is toxic to HR-dependent DNA
synthesis, in the absence of Nup132 and STUbL activity, suggesting
a role for NPCs in counteracting this toxicity.
To gauge the type of SUMOylation involved in relocation but
becoming toxic to HR-mediated DNA synthesis, we manipulated
the level and type of SUMO conjugates by several means. We
employed a
“Low SUMO” strain in which the endogenous SUMO
Fig. 3 Slx8 STUbL is necessary for anchoring to NPCs and RDR but not for safeguarding fork-integrity. a Co-localization events in S-phase cells in
indicated conditions and strains, as described on Fig.2a. p value was calculated by Fisher’s exact test. b MSD of the RFB in OFF and ON conditions in n
S phase cells of slx8-29 mutant grown at permissive (25oC, left panel) and restrictive (32 °C, right panel) temperature over indicated time interval (Δt).
p value was calculated as a one sided t-test based on MSD curves. Black bars correspond to SEM. c Diagram of constructs containing the reporter gene ura4-sd20 (green) associated (t-ura4sd20 < ori) or not (t-ura4sd20-ori) to the RFB. The non-functional ura4-sd20 allele, containing a 20-nt duplication flanked by micro-homology, is located downstream of the RFB. Upon activation of the RFB, a restarted fork can replicate the ura4-sd20 and the
HR-mediated non-processive DNA synthesis favors the deletion of the duplication, resulting in a functional ura4+gene, generating Ura+cells. As control, the
construct devoid of RFB is used to monitor the spontaneous frequency of RS that is then subtracted to obtain the frequency of RFB-induced RS.d Frequency
of RFB-induced Ura+reversion in indicated strains and conditions. Each dot represents one sample from n independent biological replicate. Bars indicate
mean values ± SD. p value was calculated by two-sided t-test. e Binding of Rad51 to the RFB in WT and slx8-29 strains at indicated temperature. ChIP-qPCR results are presented as RFB ON/OFF ratio for each mutant. Distances from the RFB are presented in bp. Values are mean from three independent
biological replicates ± SEM.f Top panel: Scheme of replication intermediates (RI) analyzed by neutral-neutral 2DGE of the AseI restriction fragment in RFB
OFF and ON conditions. Partial restriction digestion caused by psoralen-crosslinks results in a secondary arc indicated on scheme by blue dashed lines. Bottom panels: Representative RI analysis in indicated strains and conditions. The ura4 gene was used as probe. Numbers indicate the percentage of forks
promoter was replaced by a weaker constitutive promoter
53and a
pmt3-KallR mutant (SUMO-KallR) in which all internal Lys are
mutated to Arg to prevent poly-SUMOylation
55. Pli1-dependent
SUMO chain formation is enhanced by the interaction between
the single E2 SUMO conjugating enzyme Ubc9 and SUMO. Thus,
we took advantage of the pmt3-D81R mutant (SUMO-D81R) that
impairs Ubc9-SUMO interaction and allows mono and
di-SUMOylation to occur in a Pli1-dependent manner but impairs
the chain-propagating role of Pli1 that is toxic in the absence of
STUbL
55. In all conditions, the active RFB did not shift to the NP
and RFB-induced RS was slightly increased (Fig.
5
f, g), indicating
that poly-SUMOylation is instrumental in relocating the RFB but
impedes HR-dependent DNA synthesis. Moreover, all conditions
restored RFB-induced RS in nup132Δ cells, indicating SUMO
chains are the source of toxicity to RDR (Fig.
5
g). Hence,
relocation requires Pli1-dependent SUMO chain formation which
then limits HR-mediated DNA synthesis, generating a need to
overcome this inhibitory effect by events occurring at NPCs. In
addition, limiting the SUMO chain-propagating role of Pli1 is
sufficient to bypass the necessity for relocation to NPCs to ensure
efficient RDR.
Relocation to NPCs allows SUMO chains removal by Ulp1 and
the proteasome. Relocation to NPCs is necessary to overcome the
inhibitory effect of SUMO chains when priming HR-mediated
DNA synthesis. STUbLs promote the ubiquitylation of SUMO
conjugates for proteolysis by the proteasome, whose activity is
enriched at the NP
33. We focused on Rpn10, a regulatory subunit
of the proteasome, whose absence results in defective degradation
of ubiquitinated proteins
58. In rpn10Δ cells, the active RFB shifted
to the NP but the frequency of RFB-induced RS was severally
80 Controla
b
e
d
c
0.0075 % MMS 7.5 μM CPT 0.5 μg/mL Bleo 200 J/m2 UV 3 mM HU 8 7 6 5 4 3 2 1 0 8 7 6 5 4 3 2 1 0 60 40 20 0 Frequency of RFB-Induced Ura + reversion × 10 -5 80 70 60 50 40 Co-localization in S-phase %Rad51 enrichment (ratio ON/OFF)
30 20 10 0 WT WT WT WT nup131 nup131 nup132 nup132 nup131 nup131Δ nup132 nup132Δ nup132Δ WT nup131Δ nup132Δ nup131 132 nup131132 nup131 132 nup131Δ132Δ WT nup131Δ nup132Δ nup131Δ132Δ WT nup131Δ nup132Δ nup131Δ132Δ nup131Δ132Δ nup131Δ 132Δ 2.5 2.0 Fold enrichment 1.5 1.0 0.5 0.0 2.5 2.0 1.5 1.0 0.5 0.0 2.5 2.0 1.5 1.0 0.5 0.0 ade6 control locus n = 4 n = 4 n = 3 n = 3 p < 0.0001 p = 0.0001 p = 0.001 p = 0.0225p = 0.0196 p = 0.0041p = 0.0008 p = 0.0009 p = 0.0012 p = 0.0025 ns ns ns RFB ON RFB ON RFB OFF RFB OFF no RFB n = 3 n = 3WT n = 4 RFB RFB –110 110 400 –110 110 400 bp bp
–110 110 400 ade6 –110 110 400 ade6 –110 110 400 ade6
Fig. 4 Nup132 promotes HR-mediated DNA synthesis, downstream of Rad51 binding, in a post-anchoring manner. a Sensitivity of indicated strains to indicated genotoxic drugs. Ten-fold serial dilution of exponential cultures were dropped on appropriate plates. Bleo bleomycin; CPT camptothecin; HU hydroxyurea; MMS methyl methane sulfonate and UV: Ultra Violet-C. See supplementary Fig. 4 for the characterization of macromolecules transport and
supplementary Fig. 5 for replication defect upon HU-fork stalling.b Frequency of RFB-induced Ura+reversion in indicated strains and conditions. Each dot
represents one sample from seven independent biological replicate for each strain. Bars indicate mean values ± SD. p value was calculated by two-sided
t-test. c Binding of Rad51 to the RFB in indicated strains as described on Fig.3e. Values are mean from n independent biological replicates ± SEM.d
Co-localization event in S-phase cells in indicated conditions and strains. In all, 250 cells were analyzed for each condition and strain. p value was calculated by
Fisher’s exact test for OFF and ON groups for each mutant and condition. Dots represent values obtained from two independent biological experiments. For
each set of data, WT strain was analyzed alongside mutants. e Binding of Npp106-GFP to the RFB in indicated strains. Upstream and downstream distances from the RFB are presented in bp. Primers targeting ade6 gene were used as unrelated control locus. Values are mean of n independent biological repeats ± SD. p value was calculated using two-sided t-test.
2.5 2.0 Fold enrichment 1.5 1.0 0.5 0.0 2.5 2.0 Fold enrichment 1.5 1.0 0.5 0.0 2.5 2.0 Fold enrichment 1.5 1.0 0.5 0.0 RFB ON RFB OFF n = 3 n = 3 n = 3 WT –110 110 400 ade6 –110 110 400 ade6 –110 110 400 bp ade6 80 70 60 50 40 Co-localization in S-phase % pli1Δ S phase 30 20 10 0 Control
a
b
d
e
f
g
c
3 mM HU 7.5 μM CPT 0.006 % MMS WT WT WT 100 80 60 40 20 0 pli1Δ pli1 pli1Δ pli1Δ nup132Δ nup132Δ pli1Δ nup132Δ WT WT WTLow SUMO Low SUMOnup132
SUMO-D81R SUMO-D81Rnup132
SUMO-KallRnup132 SUMO-KallR nup132 Low SUMO SUMO -D81R SUMO -KallR
pli1Δ pli1Δ pli1Δ
slx8-29 pli1Δ slx8-29 slx8 -29 WT pli1Δ slx8 -29 pli1Δ nup132Δ nup132Δ pli1 nup132 RFB ON RFB OFF no RFB RFB ON RFB OFF no RFB p <0.0001 p = 0.0207 p = 0.0005 p = 0.0001 p = 0.0064 p = 0.0059 p = 0.0002 p = 0.0009 p = 0.0007 p < 0.0001 p = 0.0034 p < 0.0001 p < 0.0001 p = 0.0038 p = 0.0047 p = 0.0008 RFB OFF Rc = 0.58 μm n = 19 RFB ON Rc = 0.58 μm n = 18 0.6 0.5 MSD ( μ m 2) 0.4 0.3 0.2 0.1 0 0 50 100 150 200Δt (s) Frequency of RFB-Induced Ura + reversion × 10 -5 Frequency of RFB-Induced Ura + reversion × 10 -5 80 100 60 40 250 200 150 100 50 0 20 0 ns ns ns ns ns ns ns ns ns 25°C 32°C 80 70 60 50 40 Co-localization in S-phase % 30 20 10 0 n = 8 n = 8 n = 16 n = 12 n = 8 n = 8 n = 12 n = 12
Fig. 5 Pli1-dependent SUMO chain promotes relocation to NPCs but are toxic to RDR. a Sensitivity of strains to indicated genotoxic drugs. Ten-fold serial
dilution of exponential cultures were dropped on appropriate plates as described in Fig.4a.b Co-localization event in S-phase cells in indicated conditions
and strains as on Fig.2a. p value was calculated by Fisher’s exact test for OFF and ON groups for each mutant and condition. c Binding of Npp106-GFP to
the RFB in indicated strains. Upstream and downstream distances from the RFB are presented in bp. Primers targeting ade6 gene were used as unrelated control locus. Values are mean of three independent biological repeats ± SD. p value was calculated using two-sided t-test. d MSD of the RFB in OFF and
ON conditions in S phase cells of pli1Δ mutant over indicated time interval (Δt) calculated for n independent cells, as described on Fig.1d. Black bars
correspond to SEM.e Frequency of RFB-induced Ura+reversion in indicated strains and conditions. Each dot represents one sample from eight
independent biological replicate for each strain. Bars indicate mean values ± SD. p value was calculated by two-sided t-test. f Co-localization event in
S-phase cells in indicated conditions and strains as in Fig.2a. p value was calculated by Fisher’s exact test for OFF and ON groups for each mutant and
condition.g Frequency of RFB-induced Ura+reversion in indicated strains and conditions. Each dot represents one sample from n independent biological
decreased and a slight additivity was observed in nup132Δ rpn10Δ
cells (Fig.
6
a, b). Thus, the proteasome activity is necessary for
efficient RDR but this might not be under regulation by Nup132.
In the absence of Nup132, Ulp1 is delocalized from NPCs
that are no longer able to counteract the toxicity of SUMO
chains to promote RDR. Thus, we investigated the role of Ulp1
in RDR. The overexpression of Ulp1 rescued the defective
RFB-induced RS of nup132Δ cells (Fig.
6
b), indicating that low Ulp1
expression is detrimental to efficient RDR. We employed a
LexA-based tethering approach to artificially target Ulp1 to the
RFB
23(Fig.
6
c). Expression of Ulp1-LexA did not lead to
sensitivity to genotoxic agents in striking contrast to ulp1Δ cells
(Fig.
6
d), indicating the fusion protein is functional. Ulp1-LexA
was enriched in the vicinity of the RFB only in the presence of 8
LexA binding sites (at the t-LexBS-ura4sd20 < ori construct,
Fig.
6
e). Consistent with the role of Nup132 in anchoring Ulp1
at the NP, the inactive RFB shifted to the NP, in a Nup132
manner. When activated, the RFB shifted to the NP in the
absence of Nup132, confirming that Ulp1 is not necessary for
anchorage (Fig.
6
f). Remarkably, tethering Ulp1-LexA to the
active RFB, anchored to NPCs, resulted in an increased
frequency of RFB-induced RS in the absence of Nup132,
reinforcing the notion that Ulp1-associated NPCs are required
to overcome the inhibitory effect of poly-SUMOylation on
HR-mediated DNA synthesis (Fig.
6
g).
Pli1 safeguards the integrity of nascent strands at arrested
forks. A question arising from our work is the positive effect of
Pli1 activity at sites of replication stress. Although pli1Δ cells were
insensitive to replication-blocking agents, they exhibited a clear
defect in the recovery from HU-stalled forks and in chromosomes
duplication, suggesting an accumulation of unresolved replication
intermediates (Supplementary Fig. 5). We thus investigated the
integrity of the fork arrested at the RFB by 2DGE and observed an
increased level of resected forks in pli1Δ cells (Fig.
7
a, b).
RPA-ChIP confirmed an extensive recruitment of RPA, up to 3 Kb
upstream of the RFB, supporting the formation of larger ssDNA
gaps in the absence of Pli1 (Fig.
7
c). Thus, Pli1 activity is critical
to negatively regulate the resection of nascent strands and
safe-guard fork-integrity.
Discussion
Collapsed forks anchor to NPCs but the mechanisms engaged at
NPCs to ensure fork integrity and restart were not understood.
Here, we reveal the beneficial and detrimental functions of
SUMOylation at replication stress sites. We propose that Pli1
activity engages at arrested forks to control the extent of nascent
strand resection. Pli1 generates SUMO chains that signal for a
STUbL-dependent anchorage to NPCs, but hinder the priming of
HR-mediated DNA synthesis. Hence, NPCs become critical to
allow the resumption of DNA synthesis by clearing off SUMO
conjugates in a post-anchoring manner, via Ulp1 and proteasome
activities. Selectively preventing Pli1-mediated SUMO chains
bypasses the need for anchorage to NPCs while maintaining
efficient RDR. Thus, SUMO-regulated mechanisms spatially
segregate the subsequent steps of RDR from Rad51 loading and
activity occurring in the nucleoplasm and the restart of DNA
synthesis occurring after anchorage to NPCs (Fig.
7
d).
We establish that DSB formation is not a requirement to
anchor arrested forks to NPCs. Instead, it requires forks to be
remodeled by Rad51 enzymatic activity. Relocation requires
nascent strand resection to occur for Rad51 loading, but is not
sufficient per se. SUMOylation of HR factors is necessary to
anchor expanded CAG tracts to NPCs
59and therefore their
absence at the RFB may impair the wave of SUMOylation
necessary for relocation. However, the lack of relocation in the
Rad51-II3A mutant indicates that joint-molecules, such as
D-loops from which DNA synthesis is primed, are also relevant
positioning signals to relocate arrested forks to NPCs. In several
eukaryotes, relocation of DSBs to the NP requires end-resection
and Rad51, suggesting that Rad51-mediated repair progression
stabilizes repair intermediates to facilitate anchorage
59. Breaks
within repeated sequences (heterochromatin in
flies, mouse
peri-centromere, rDNA in budding yeast) shift away from their
compartments to continue HR repair and load Rad51 at
mobi-lized DNA damage sites
26,43,45,60. Relocation of forks collapsed at
expanded CAG repeats requires nuclease activities to engage
SUMO-RPA onto ssDNA which prevents Rad51 loading.
Anchorage to NPCs then facilitates Rad51 loading
59. Here, we
report a distinct situation when forks arrest within a unique
sequence. Relocation requires Rad51 loading and enzymatic
activity and the lack of anchorage (in STUbL or nucleoporin
mutants) does not affect Rad51 loading, supporting that Rad51
loading and enzymatic activity occur prior to anchorage to NPCs.
These distinct situations likely reflect different mechanisms
engaged at unique sequence versus repeated sequences, where
controlling Rad51 loading is of major importance to avoid
potential rearrangements for the latter.
STUbL binds to SUMO modified DNA repair factors via its
SIM domains to tether DNA lesions to NPCs
16,59. Our data are
consistent with this and highlight the positive and negative effects
of bulk SUMOylation mediated by Pli1. Though the potential
mode of Pli1 recruitment to replication stress sites remain to be
identified, we show that Pli1 engagement at arrested forks is vital
to safeguard fork-integrity. We noticed that the lack of Pli1 did
not increase RDR efficiency whereas preventing SUMO chains
does, suggesting that Pli1-dependent mono-SUMOylation events
remain necessary to RDR. The Ubc9-SUMO interface may help
to increase the local concentration of SUMO particles to enhance
Pli1-mediated SUMO chains and mediate anchorage to NPCs. In
contrast to forks collapsed at CAG tracts
59, relocation requires
poly-SUMOylation as reported for persistent DSBs in budding
yeast
20. However, those SUMO chains limit HR-mediated DNA
synthesis, possibly the DNA synthesis primed from D-loops, a
step necessary to ensure efficient fork restart. A selective defect in
Pli1-mediated SUMO chain or preventing poly-SUMOylation
bypasses the need for relocation to NPCs and alleviates the
toxicity of SUMO conjugates. A remaining question is whether
the SUMO-targets responsible for relocation and preventing the
priming of HR-mediated DNA synthesis are similar or distinct.
A possible scenario is that SUMO-dependent relocation to
NPCs occurs when arrested forks are not rescued in a timely
manner by opposite forks: this would lead to safeguarding
fork-integrity by Pli1, and thus engaging the relocation process to
NPCs. Interestingly, the lack of STUbL resulted in increased
mobility of arrested forks, a phenomena not observed in the
absence of Pli1, suggesting that SUMOylation promotes
chro-matin mobility of replication stress sites and STUbL promotes
their anchorage to NPCs.
Collectively, this study uncovers how anchorage to NPCs helps
to sustain DNA synthesis upon replication stress. The lack of
Nup132 provides a unique genetic situation to uncouple the role
of NPCs in anchoring arrested forks from their role in promoting
DNA synthesis upon stress conditions. We establish that Nup132
is necessary to prime HR-mediated DNA synthesis, downstream
of Rad51 binding and activity, in a post-anchoring manner. This
function is linked to the role of Nup132 in recruiting Ulp1 at
NPCs and is uncoupled from the transport of macromolecules.
We propose that Ulp1-associated NPCs, as well as proteasome
activity, are critical to remove SUMO conjugates from
joint-molecules to allow DNA synthesis resumption. Consistent with
80
a
c
e
d
f
g
b
70 60 50 40 Co-localization in S-phase % 30 20 10 0 RFB ON RFB OFF RFB ON RFB OFF no RFB Frequency of RFB-Induced Ura + reversion X 10 -5 150 100 50 0 200 10 % IP/INPUT 8 6 4 2 0 150 100 50 0 WTWT rpn10Δ nup132Δrpn10Δ rpn10Δrpn132Δ WT nup132Δ pnmt81-ulp1pnmt81-ulp1nup132Δ
p < 0.0001 p = 0.0028 p = 0.0001 p = 0.0005 p = 0.0046 p = 0.0004 p = 0.0004 p = 0.0001 p = 0.0003 p = 0.041 ns t-lexBS-ura4sd20 <ori t-ura4sd20 <ori : ura4sd20 ulp1-lexA lexBS RFB
ori ori ori ori
cen3 0.5 μM CPT Control 1 mM HU Ulp1 Ulp1 – + – + Ulp1 Ulp1-lexA Ulp1-lexA Ulp1-lexA 3 mM HU 7.5 μM CPT WT ulp1Δ Ulp1-lexA Ulp1-lexA t-lesBS-ura4sd20 <ori t-LacO-ura4::lexBS <ori lexBS at WT ulp1Δ Ulp1-lexA
Ulp1-lexA t-ura4sd20 <ori Ulp1-lexA t-lesBS-ura4sd20 <ori
Ulp1-lexA t-lesBS-ura4sd20 <ori WT WT ura4 ade6 (control locus) nup132 nup132 nup132 WT nup132 Frequency of RFB-Induced Ura + reversion × 10 -5 80 100 60 40 20 0 80 70 60 50 40 Co-localization in S-phase % 30 20 10 0 n = 4 p = 0.0003 p = 0.0029 p = 0.0002 p = 0.0007 p = 0.0003 p < 0.0001 ns
Fig. 6 Proteasome and Ulp1 activity are necessary to clear off SUMO conjugates to promote RDR. a Co-localization event in S-phase cells in indicated
conditions and strains as on Fig.2a. p value was calculated by Fisher’s exact test for OFF and ON groups for each mutant and condition. b Frequency of
RFB-induced Ura+reversion in indicated strains and conditions. Each dot represents one sample from eight independent biological replicate for each strain.
Bars indicate mean values ± SD. p value was calculated by two-sided t-test. c Diagram of construct containing lexA-binding site (lexBS, purple) that allows tethering of Ulp1-lexA to the t-lexBS-ura4sd20 < ori construct (d, e, g) or to the t-Laco-ura4::lexBS < ori construct (f). d Sensitivity of indicated
strains to indicated genotoxic drugs. Ten-fold serial dilution of exponential cultures were dropped on appropriate plates.e Binding of Ulp1-LexA to ura4 or
ade6 (unrelated control locus) in the presence of LexBS (t-lexBS-ura4sd20 < ori) or not (t-ura4sd20 < ori). Values are mean of four independent biological
repeats ± SD. p value was calculated using two-sided t-test. f Co-localization event in S-phase cells in indicated conditions and strains as on Fig.2a. p value
was calculated by Fisher’s exact test for OFF and ON groups for each mutant and condition. g Frequency of RFB-induced Ura+reversion in indicated strains
and conditions. Each dot represents one sample from 14 independent biological replicate for each strain. Bars indicate mean values ± SD. p value was calculated by two-sided t-test.
budding yeast Nup84 sustaining fork progression at stalled forks
41,
Nup132 is necessary to sustain DNA replication upon HU
treat-ment. The deletion of Pli1 did not rescue the defect in the recovery
from HU-stalled forks in nup132Δ cells (Supplementary Fig. 5),
indicating that Nup132 sustains DNA replication upon stress by
distinct mechanisms according to the nature of stalled versus
dysfunctional forks.
SUMOylation is a dynamic and reversible modification. At
dysfunctional forks, our data establish a clear role of NPCs in
counteracting the toxicity of SUMO chains to allow HR-mediated
DNA synthesis. SUMO removal involves Ulp1 and the
protea-some, two activities occurring at the NP. Although the role of
NPCs in promoting the removal of SUMO conjugates has been
previously proposed
17, our work reveals the versatile functions of
SUMOylation in promoting fork integrity and relocation at the
expense of limiting the step of HR-mediated DNA synthesis. We
propose that SUMO-primed ubiquitylation promotes the
clear-ance of DNA repair/replication factors at arrested forks to prime
DNA synthesis, but the multiple targets remain unknown.
Interestingly, the Branzei lab recently identified replication factors
undergoing SUMOylation regulated by Ulp2 and STUbL to
control replication initiation
61. Similarly, we propose that key
SUMOylated factors are controlled by Ulp1 and STUbL to
reg-ulate timely fork restart.
Methods
Standard yeast genetics. Yeast strains and primers used in this work are listed in Table S1 and S2 respectively. Gene deletion or tagging were performed by classical genetic techniques. Strain with SUMO-KallR was obtained by integra-tion of synthetized mutated pmt3 gene (Genscript) into pmt3::ura4 and colonies
were selected on 5-FOA. Mutation of all lysines to arginines was confirmed by
sequencing. To assess the sensitivity of chosen mutants to genotoxic agents, midlog-phase cells were serially diluted and spotted onto plates containing hydroxyurea (HU), methyl methanesulfonate (MMS), campthotecin (CPT), bleomycin (bleo) or irradiated with an appropriate dose of UV. Strains carrying the RTS1, replication fork block sequence were grown in minimal medium EMMg (with glutamate as nitrogen source) with addition of appropriate sup-plements and 60 µM thiamine (barrier inactive, OFF). The induction of repli-cation fork block was obtained by washing away the thiamine and further incubation in fresh medium for 24 h (barrier active, ON).
Live cell imaging. For snapshot microscopy, cells were grown infiltered EMMg
with or without 60 µM thiamine for 24 h to exponential phase (RFB OFF and RFB ON), then centrifuged and resuspended in 500 µL of fresh EMMg. In all, 1 µL from resulting solution was dropped onto Thermo Scientific slide (ER-201B-CE24) covered with a thin layer of 1.4% agarose infiltered EMMg. 21 z-stack pictures (each z step of 200 nm) were captured using 3D LEICA DMRXA microscope, supplied with CoolSNAP monochromic camera (Roper Scientific) under 100X
oil-immersion magnification with numerical aperture 1.4. Exposure time for GFP
channel was 500 ms, for mCherry 1000 ms. Pictures were collected with META-MORPH software and analyzed with ImageJ software. Foci that merged or partially overlap were counted as colocalization event.
The mobility of arrested forks was investigated by collecting 3-dimensional 14-stack images every 1.5 s over 5 min. Cells were visualized with a Spinning Disk Nikon inverted microscope equipped with the Perfect Focus System, Yokogawa CSUX1 confocal unit, Photometrics Evolve512 EM-CCD camera, 100X/1.45-NA PlanApo oil immersion objective and a laser bench (Errol) with 491 diode laser, 100 mX (Cobolt). Images were captured every 1.5 s with 14 optical slices (each z step of 300 nm), 100 ms exposure time for single GFP channel at 15% of laser power using METAMORPH software. Time-lapse movies were mounted and analyzed with ImageJ software as described below.
To study the colocalization time between lacO/LacI RFB foci and Npp106-GFP cells grown in the above conditions were visualized with a Nikon inverted
microscope described above, using twofluorescent channels with 491 and
561 nm diode lasers, 100 mX (Cobolt). Images were captured every 10 s with 14 optical slices (each z step of 300 nm) for 30 min with 100 ms exposure time both for GFP and mCherry channels at 15% of laser power using METAMORPH software. Time-lapse movies were mounted and analyzed with ImageJ software (description below).
Protein import-export from nucleus was monitored using WT and
nup131Δnup132Δ strains expressing genomic LacI-NLS-GFP without LacO repeats integrated into the genome. Cells grown for 24 h with or without thiamine were visualized with Nikon inverted microscope described above. Snapshot pictures (21 stacks, each z of 200 nm and 100 ms exposure) were acquired using METAMORPH software and analyzed in ImageJ. Images were projected for maximum intensity. The nuclear/cytoplasmic ratio (N/C) was determined by
measuring meanfluorescence intensity within constant square regions (ROI plugin
from ImageJ) placed in the cytoplasm, center of nucleus and intercellular background. Nuclear/cytoplasm ratio stand for (Nucleus-background)/ (Cytoplasm-background).
All image acquisition was performed on the PICT-IBiSA Orsay Imaging facility of Institut Curie.
Movie analysis. Movies have been mounted using ImageJ. For analysis of mobility of arrested forks after projection around z-axis, single-particle tracking was per-formed using ImageJ plugin SpotTracker62. Obtained coordinates for RFB foci were 95.4 ± 2.1 96.3 ± 1.4
c
b
WT pli1Δ RFB ON RFB OFFa
Resected forks %(relative to arrested forks)
WT pli1Δ 0 5 10 15 20 25 30
d
0 0.5 1 1.5 2 2.5 ade6 control locus 0 0.5 1 1.5 2 2.5 -110 110 400 600 900 1400 2200 3000 WT pli1Δ RPA enrichment (ratio ON/OFF)RFB RFB bp p = 0.0003 p = 0.0028 p = 0.0034 n = 4 n = 4 WT pli1Δ n = 4 n = 4 Nucleus Cytoplasm Rad51 Rad51 activity Mono-SUMO SUMO SUMO Slx8 Rfp1/2 Nup131 Nup132 Nuclear pore Ulp1 Proteasome SUMO chains removal Priming HR-mediated DNA synthesis Anchorage pli1
Fig. 7 Pli1 safeguards fork-integrity by limiting resection of nascent strands. a Representative RI analysis in indicated strains and conditions as
described on Fig.3.b Quantification of resected forks. Values are mean of
four independent biological replicates ±SD. p value was calculated by two-sided t-test. c Binding of RPA (Ssb3-YFP) to the RFB in indicated strains. ChIP-qPCR results are presented as ON/OFF ratio for each mutant. Upstream and downstream distances from the RFB are presented in bp. Values are mean from four independent biological replicates ±SD. p value was calculated by two-sided t-test. Primers targeting ade6 gene were used
as unrelated control locus.d SUMO-based regulation of relocation of
